The Molecular Mechanism of Ginsenoside Analogs Activating TMEM16A

Abstract

The calcium-activated chloride channel TMEM16A is involved in many physiological processes, and insufficient function of TMEM16A may lead to the occurrence of various diseases. Therefore, TMEM16A activators are supposed to be potentially useful for treatment of TMEM16A downregulation-inducing diseases. However, the TMEM16A activators are relatively rare, and the underlying activation mechanism of them is unclear. In the previous work, we have proved that ginsenoside Rb1 is a TMEM16A activator. In this work, we explored the activation mechanism of ginsenoside analogs on TMEM16A through analyzing the interactions between six ginsenoside analogs and TMEM16A. We identified GRg2 and GRf can directly activate TMEM16A by screening five novel ginsenosids analogs (GRb2, GRf, GRg2, GRh2, and NGR1). Isolated guinea pig ileum assay showed both GRg2 and GRf increased the amplitude and frequency of ileum contractions. We explored the molecular mechanisms of ginsenosides activating TMEM16A by combining molecular simulation with electrophysiological experiments. We proposed a TMEM16A activation process model based on the results, in which A697 on TM7 and L746 on TM8 bind to the isobutenyl of ginsenosides through hydrophobic interaction to fix the spatial location of ginsenosides. N650 on TM6 and E705 on TM7 bind to ginsenosides through electrostatic interaction, which causes the inner half of α-helix 6 to form physical contact with ginsenosides and leads to the pore opening. It should be emphasized that TMEM16A can be activated by ginsenosides only when both the above two conditions are satisfied. This is the first, to our knowledge, report of TMEM16A opening process activated by small-molecule activators. The mechanism of ginsenosides activating TMEM16A will provide important clues for TMEM16A gating mechanism and for new TMEM16A activators screening.

Significance

TMEM16A is one of the most important anion channels in the human body; however, its gating mechanism is still unclear. In this work, we investigated the activation of TMEM16A by a class of ginsenoside analogs. We proposed the molecular mechanism of TMEM16A activation by ginsenosides through analyzing the interaction between different ginsenoside analogs and TMEM16A. This is the first, to our knowledge, activation model of TMEM16A activated by small-molecule activators. The mechanism of ginsenosides activating TMEM16A will provide important clues for TMEM16A gating mechanism research and new TMEM16A activators screening.

Introduction

Calcium-activated chloride channels (CaCCs) are important anion channels that are widely distributed in various tissues of the human body and are involved in a variety of physiological processes (1). Since TMEM16A was confirmed as the molecular basis of CaCCs in 2008 (2, 3, 4), many research groups began to study the structure and function of TMEM16A. Growing evidence has demonstrated that abnormal expression of TMEM16A is closely related to asthma, hypertension, diarrhea, gastrointestinal dysfunction, pain, and some cancers (5, 6, 7). At the end of 2017, two groups independently reported the cryogenic electron microscopy (cryoEM) structure of the TMEM16A protein (8,9). The research groups elaborated on the mechanism of the calcium activation process of TMEM16A and proposed a double-barreled model of TMEM16A. The appearance of the cryoEM structure greatly promoted the study of TMEM16A and helped us to explore the regulatory mechanism of the TMEM16A modulators. In addition, with the help of the cryoEM structure, we can explore the gating mechanism of TMEM16A by studying the regulation of TMEM16A by modulators.

Gastrointestinal dysfunction is one of the diseases associated with downregulated TMEM16A function (10,11). TMEM16A protein is expressed in the interstitial cells of Cajal (ICCs), which are the pacemaker cells for smooth muscle contractions in the gastrointestinal tract (12). Studies have shown that the gastrointestinal muscles cannot form slow waves when TMEM16A is knocked out in mouse ICCs (13). Therefore, the activation of TMEM16A in the gastrointestinal tract can be used for gastrointestinal dysfunction therapy (11, 38). It has been one of the important means to reflect the activity of TMEM16A by observing the change of ileal tension in guinea pigs.

In our previous study, we found that GRb1 (ginsenoside Rb1) is a safe natural TMEM16A activator that can enhance guinea pig ileal contractions (11). Ginsenosides are the principal ingredients of ginseng (14). Ginsenosides are the active components responsible for the pharmacological properties of ginseng, including its antidiabetic, vasorelaxant, antineoplastic, antiinflammatory, and antioxidant activities (15,16). There are also some studies showing that ginsenosides play critical roles in promoting gastrointestinal motility (17,18). However, most of the pharmacological effects have not been associated with a certain receptor on the cell membrane. The major components of total ginsenosides are ginsenosides Rb1, Rb2, Rc, Rd, Re, and Rg1, which compose more than 90% of all ginsenosides (19), whereas other diverse ginsenosides are present in relatively small amounts, including ginsenosides Rg2, Rg3, Rh2, Rh1, and Rf (20). In this study, we set to explore if any other components of ginseng can interact with TMEM16A and if possible, find common features of these ginsenoside analogs in activating TMEM16A. These results may provide some guidance for the exploration of TMEM16A gating mechanism.

In this work, we studied the activation mechanism of ginsenoside analogs on TMEM16A by analyzing the interactions between six ginsenoside analogs and TMEM16A. It turned out that among the six ginsenoside analogs (GRb1, GRb2, GRf, GRg2, GRh2, and NGR1), which have a high structural similarity, GRb1, GRg2, and GRf can directly activate TMEM16A, whereas the other three analogs can barely activate TMEM16A (Table 1). Next, we verified the enhancement effect of GRg2 and GRf on ileum contractions in guinea pigs. Then, we analyzed the interactions between GRb1, GRg2, GRf, and TMEM16A, respectively. Interestingly, the results showed that all of the above three ginsenoside analogs formed hydrogen bonds with N650 and E705 in TMEM16A and formed hydrophobic interaction with A697 and L746 in TMEM16A. Based on the spatial positions of N650, E705, A697, and L746, we proposed the mechanism of TMEM16A channel opening process activated by ginsenosides. This TMEM16A ion channel opening mechanism activated by ginsenosides will promote the exploration of TMEM16A gating mechanism.

Table 1.

The Structure of Ginsenoside Analogs

Materials and Methods

Cell preparation

All cells were maintained under standard cell culture conditions of 5% CO₂ and 95% humidity. CHO cells were cultured in F-12K medium (Gibco, Waltham, MA) with 10% fetal bovine serum (Sijiqing, Hangzhou, China), 100 IU/mL penicillin, and 100 μg/mL streptomycin (Beijing Solarbio Science and Technology, Beijing, China). The F-12K medium for HEK293 cell culture was replaced by DMEM (Dulbecco's modified Eagle medium; Gibco). The mouse complementary DNA clone mANO1 (TMEM16A, accession number NM_178642.5) was kindly provided by Professor Young Duk Yang (Seoul National University, Seoul, South Korea). The stable transfection of TMEM16A into CHO cells was established according to previously published literature (21) with X-tremeGENE HP (Roche, Basel, Switzerland). The stably transfected CHO cells were seeded in 24-well plates on 14-mm glass coverslips 24 h before patch recordings. All transient transfections were performed on HEK293 cells. HEK293 cells were seeded and cultured in 24-well plates on 14-mm glass for 12 h, followed by an addition of 50 μL Opti-MEM (Solarbio), 1.5 μL X-tremeGENE HP, and 0.5 μg plasmids to the medium after shaking and were allowed to stand for 15 min.

Solutions

GRg2 (CAS no. 52286-74-5), GRf (CAS no. 52286-58-5), GRb2 (CAS no. 11021-13-9), GRh2 (CAS no. 78214-33-2), and NGR1 (CAS no. 80418-24-2) were purchased from Solarbio. All of these compounds were dissolved in DMSO (dimethyl sulfoxide) to a final concentration of 10 mM and stored at −20°C. Then, the stock solutions were diluted to a working solution using a D-PBS buffer. The D-PBS buffer contained 2.67 mM KCl, 1.47 mM KH₂PO₄, 138 mM NaCl, and 8.1 mM Na₂HPO₄ and was adjusted to pH 7.4 with NaOH. Tyrode’s solution contained 8.0 g/L NaCl, 0.2 g/L KCl, 0.26 g/L MgSO₄ ⋅ 7H₂O, 0.065 g/L NaH₂PO₄ ⋅ 2H₂O, 1.0 g/L NaHCO_3, 0.2 g/L CaCl₂, and 1.0 g/L glucose.

In the whole-cell patch clamp, the pipette solution containing 130 mM CsCl, 10 mM EGTA, 1 mM Mg ATP, 1 mM MgCl₂ ⋅ 6H₂O, and 10 mM HEPES was adjusted to pH 7.4 with CsOH. The bath solution containing 150 mM NaCl, 1 mM MgCl₂ ⋅ 6H₂O, 10 mM HEPES, 10 mM glucose, and 10 mM mannitol was adjusted to pH 7.4 with NaOH. Standard CaCl₂ (1 M; Sigma-Aldrich, St. Louis, MO) was added to produce various free Ca²⁺ concentrations; these concentrations were calculated using the CaEGTA Calculator version 1.2, available online at http://www.stanford.edu/∼cpatton/CaEGTA-NIST.htm. The bath solution containing 600 nM free Ca²⁺ was prepared by adding the standard CaCl₂ solution to a final concentration of 8.69 mM and adjusting the pH to 7.4 with CsOH. The osmotic pressure of the pipette solution was in the range of 290–300 mOsm/L and that of the bath solution was 300–310 mOsm/L, as measured by an OM815 osmometer (Löser Messtechnik, Berlin, Germany).

In the inside-out patch clamp, the pipette solution contained 130 mM NaCl, 1 mM MgCl₂ ⋅ 6H₂O, and 10 mM HEPES and was adjusted to pH 7.4 with NaOH. The bath solution contained 140 mM NaCl, 1 mM MgCl₂ ⋅ 6H₂O, 10 mM HEPES, and 5 mM EGTA, adjusted to pH 7.4 with NaOH. The free Ca²⁺ concentrations were calculated using the web server as mentioned above.

Fluorescence assay

Confocal laser scanning microscopy (CLSM) (Leica SP5, Leica Microsystems, Wetzlar, Germany) was used to the Premo Halide Sensor fluorescence assay (Thermo Fisher Scientific, Waltham, MA) for screening the TMEM16A channel activators. The YFP-F46L/H148Q/I152L plasmid was transfected into stably transfected TMEM16A CHO cells for 36 h before the fluorescence experiment. The GRg2 and GRf working solutions were added to the 24-well plate at a final concentration of 100 μM and incubated for 30 min. The test cells in the 24-well plates were washed three times with D-PBS and 500 μL D-PBS was left in each well. The 24-well plate was then placed on the objective table for fluorescence observation. The YFP-F46L/H148Q/I152L was excited with a 488-nm beam from a Kr/Ar laser, and emissions were detected with a standard fluorescein filter set at 520 ± 15 nm. When the fluorescence intensity curve stabilized, 500 μL iodide containing D-PBS buffer (150 mM I⁻) was added to each well. 200 photos were continuously taken in the fluorescence detection experiment with xy-t mode of CLSM (a live photo was taken every 1.25 s).

Electrophysiology

All patch-clamp experiments were performed at room temperature (22–25°C). Patch pipettes were pulled from borosilicate glass capillaries with an outer diameter of 1.5 mm and an inner diameter of 0.86 mm (Sutter Instrument, Novato, CA) using a P-97 puller (Sutter) with a pipette resistance of 3–5 MΩ (whole-cell mode) or 1–1.5 MΩ (inside-out mode) when immersed in the bath solution. The seal resistance of the patch clamp was typically 1–2 GΩ or higher. Data recordings were performed with an EPC10 amplifier controlled by Pulse software with a Digi LIH1600 interface (HEKA Elektronik, Lambrecht, Germany). The data were low-pass filtered at 2.9 kHz and sampled at 10 kHz. The stimulation protocol consisted of voltage steps of 1150 ms durations from a holding potential of 0 mV for 100 ms; the membrane voltage was clamped in steps of 20 mV from −80 to +80 mV for 750 ms, then back down to −80 mV for 300 ms.

Cytoplasmic calcium measurements

CHO cells stably transfected with TMEM16A were cultured for 24 h in 24-well plates. The cells were incubated with the fluorescent calcium indicator Fluo 3-AM (10 μM) for 30 min in the dark at 37°C before measurement. Then, the 24-well plate was placed on the objective table for observation using CLSM. The excitation wavelength of the Fluo 3-AM was 488 nm, and the emission wavelength was 515 nm. Then, 100 μM GRg2 or 200 μM GRf were added to the well when the fluorescence intensity curve stabilized, and 10 μM ATP and 10 μM A23187 were added to the well, set at ∼1.5 min intervals.

Intestinal smooth muscle contraction assay

All animal experiments were carried out in accordance with the approved guidelines of the Animal Care and Use Committee at the Hebei Medical University Experimental Animal Center (license no. SCXK (Ji) 2008-1-003; certificate no. 909106). Guinea pigs (CL grade, 300–400 g) were fasted for 24 h and killed using avertin overdose (200 mg/kg). The ileum was excised quickly and washed three times with Tyrode’s solution. The ends of the ileal segments were connected to a force transducer with silk-thread ties. Ileal segments were subjected to a resting force of 1 g and equilibrated for 60 min, and the bath solution was changed every 15 min. The tension was monitored continuously with a Panlab four-channel physiological recorder (Harvard Apparatus, Holliston, MA). Data were recorded and analyzed using the LabChart 7.0 system.

Cell viability assay

4000–7000 cells in 100 μL medium were seeded into each well of a 96-well plate and cultured under standard cell culture conditions for 24 h. Then, the cells were incubated with the control DMSO or GRg2 or GRf (100–500 μM) for 24 h. 20 μL MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) solution with 5 mg/mL concentration was added to each well and incubated for 4 h at 37°C in the dark. Then, the supernatant in the 96-well plate was discarded, and 150 μL DMSO was added to each well. The 96-well plate was agitated on a shaker at 100 rounds per minute for 10 min to fully dissolve the MTT crystals. The absorbance at 490 nm was detected by SpectraMAX i3 (Molecular Devices, San Jose, CA). The percentage of cell viability was calculated by dividing the absorbance of the test group by the absorbance of the control group.

Molecular docking and interaction analysis

The mTMEM16A structure used in the molecular simulation was obtained by homology modeling. The tertiary structure of the mTMEM16A monomer was constructed using a calcium-bound mTMEM16A chloride channel (Protein Data Bank, PDB: 5OYB) (9). The SWISS-MODEL server was used to model the mTMEM16A structure, which complements the missing structure of 5OYB (22).

We used a two-step docking strategy, which combines global random docking with local docking for molecular docking. 100 times of global random docking were performed without any restrictions. Then, a 35 × 35 × 35 Å domain was selected for local docking, which is centered on the area where ginsenosides are most distributed in the global random docking. Molecular docking of ginsenosides to mTMEM16A was performed with an AutoDock 4.2 program using the implemented empirical free energy function and the Lamarckian genetic algorithm (23). The grid maps were calculated using AutoGrid 4.2. The independent docking run performed for each docking simulation was set to 100. Cluster analysis was performed on the docked results using a root mean-square tolerance of 2.0 Å, and the initial coordinates of the ligand were mapped using ChemBioDraw Ultra 12.0. Discovery Studio Visualizer was used to analyze the hydrophobic interaction between drugs and receptors. Visualization and analysis of complexes were performed using VMD 1.9 and Pymol 0.99 (24).

Site-directed mutagenesis

Agilent Primer Design web site (https://www.agilent.com/store/primerDesignProgram.jsp) was used to design the site-directed mutagenesis primers: N650A primer, 5′-gggcaagcagctaatccaggccaatctcttcgagattggc-3′; A697S primer, 5′-cttcaacctcgaacctttcagcggcctcacgcc-3′; E705Q primer, 5′-cacgcccgagtacatgcaaatgatcattcagttc-3′; and L746T primer, 5′-caaaaagtttgtcaccgagacacggaggccagtagccatc-3′. Primer synthesis was completed by Sangon (Sangon Biotech, Shanghai, China). Fast Mutagenesis System Kit (FM111-02; Transgen, Beijing, China) was used to conduct site-directed mutagenesis with a 50 μL PCR system. The mutated plasmid was sequenced by Sangon.

Calculation of the electrostatic surface potential

The electrostatic surface potential of TMEM16A protein and ginsenoside analogs was calculated using the APBS (25) in Chimera software (26). In the proteins, missing hydrogen atoms were added, and the charges were added to all atoms before calculation. ChemBioDraw Ultra 12.0 was used for energy minimization of the ginsenoside analogs. The antechamber module of AMBER16 was used to calculate the charge of the ginsenoside analogs (27).

Data analysis

Graphical presentation and data analysis were performed using Origin 8.0. All data are presented as the mean values ± standard error. Repeated t-tests were used for comparing two groups of data. A p-value less than 0.05 was considered significant and designated by ^∗. A p-value less than 0.01 was considered very significant and designated by ^∗∗. The capacitive transients of some traces in the figures were trimmed for clarity.

Results

YFP fluorescence was quenched by the influx of I⁻ through TMEM16A activated by GRg2 and GRf

To investigate whether the ginsenoside analogs have activating effects on TMEM16A, we selected five new, to our knowledge, ginsenoside analogs for cell-based YFP fluorescent intensity detection assay. The results showed that the YFP fluorescence intensity of GRg2 and GRf incubated groups was quenched ∼90% after I⁻ was added to the bath solution (Fig. 1 A). Real-time fluorescence images in Fig. 1 B showed that the addition of DMSO (control) did not cause YFP fluorescence quenching, whereas ATP (positive control), GRg2, and GRf caused different degrees of YFP fluorescence quenching. ATP raises IP3 by regulating G-protein-coupled receptors, leading to an increase of [Ca²⁺]_i through cascade reaction (28). TMEM16A was activated by increased [Ca²⁺]_i, which resulted in I⁻ entry into the cells to quench the YFP fluorescence. Therefore, we discovered that GRg2 and GRf can activate TMEM16A, which could quench the YFP fluorescence by influx of I⁻.

Figure 1.

GRg2 and GRf quench YFP fluorescence. (A) The normalized YFP fluorescence intensity quenching curves of DMSO, ATP, GRg2, and GRf (n = 5) are shown (the bath solution is D-PBS buffer solution without Ca²⁺). (B) Representative real-time fluorescence images before and after adding DMSO, ATP, GRg2, and GRf (n = 5) are shown. To see this figure in color, go online.

GRg2 and GRf can directly activate TMEM16A in a concentration-dependent manner

Whole-cell patch-clamp experiments were performed with different concentrations of GRg2 and GRf in pipette solutions to investigate the concentration dependence of GRg2 and GRf on TMEM16A. Whole-cell patch-clamp results indicated that 1 or 10 μM of GRg2 can barely activate TMEM16A, whereas the TMEM16A current reached a maximum of 1300 pA (+80 mV) when the concentration of GRg2 increased to 100 μM (Fig. 2 A). The current density-voltage curve showed that both the outward and inward currents increased with increasing GRg2 concentration, but the increasing of the outward currents’ magnitude was more obvious than that of the inward ones (Fig. 2 B). Similar results were obtained when the pipette solutions containing more than 200 μM GRf induced a maximum of 1000 pA (+80 mV) TMEM16A current (Fig. 2, C and D). The dose-response curves of GRg2 and GRf were fitted with the Hill equation used Origin 8.0. Data fitting results showed that the EC₅₀ values of GRg2 and GRf were 32 ± 3 μM and 101 ± 4 μM, respectively. All of the results indicated that the TMEM16A channel can be activated by GRg2 and GRf in a concentration-dependent manner. In addition, the inside-out patch-clamp results showed that the EC₅₀ values of GRg2 and GRf were similar to the whole-cell patch-clamp results (Fig. S1).

Figure 2.

Concentration-dependent activation of TMEM16A by GRg2 and GRf in whole-cell patch-clamp experiments. (A) Representative currents of different concentrations of GRg2 activated TMEM16A (n = 5) are shown (both the pipette solution and bath solution are free of Ca²⁺). (B) The current density-voltage curve of different concentrations of GRg2 activated TMEM16A (n = 5) is shown. (C) Representative currents of different concentrations of GRf activated TMEM16A (n = 5) are shown. (D) The current density-voltage curve of different concentrations of GRf activated TMEM16A (n = 5) is shown. (E). The dose-response curve of GRg2 and GRf in stably transfected CHO cells with TMEM16A (n = 5) is shown. To see this figure in color, go online.

Next, we explored the activation mode and specificity of GRg2 and GRf. The Fluo 3-AM fluorescence probe was used to monitor the intracellular calcium concentration [Ca²⁺]_i after addition of GRg2 and GRg. The results showed that the intracellular fluorescence intensity did not change after adding GRg2 or GRf (Fig. S2, A and B). Then, the intracellular fluorescence intensity increased significantly and decreased with time elapsing after adding ATP. Subsequently, the Ca²⁺ ionophore A23187, which carried Ca²⁺ to the cytoplasm, increased the fluorescence intensity. Based on this observation, we determined that GRg2 and GRf did not activate TMEM16A through increasing intracellular calcium levels. On the other hand, inside-out patch clamps were adopted to verify the direct activation of TMEM16A by GRg2/GRf. Fig. S2 C showed that both the bath solution containing 100 μM GRg2 and 200 μM GRf (with trace Ca²⁺ chelated by 5 mM EGTA) can activate TMEM16A; all of the currents can be inhibited by the TMEM16A inhibitor CaCC_inh-A01. Collectively, the above experimental results showed that GRg2 and GRf can directly activate TMEM16A.

Whole-cell patch clamp was used to explore the specificity of GRg2 and GRf with four chloride channels: TMEM16A, TMEM16B, bestrophin-1, and CFTR (Fig. S3). The pipette solution containing 100 μM GRg2 activated TMEM16A and produced outward currents larger than those of 600 nM Ca²⁺, whereas 200 μM GRf produced currents equal to those of 600 nM Ca²⁺. All TMEM16A currents could be inhibited by 20 μM CaCC_inh-A01. Whole-cell patch-clamp experiments with TMEM16B showed that both GRg2 and GRf could activate TMEM16B, but the currents were smaller than those of 600 nM Ca²⁺. However, GRg2 and GRf could barely activate the bestrophin-1 and CFTR chloride channels. Therefore, we concluded that GRg2 and GRf could activate TMEM16A and TMEM16B but not bestrophin-1 and CFTR.

GRg2 and GRf enhance ileum intense contraction in guinea pigs

ICCs are pacemaker cells that control smooth muscle contractions in the stomach and intestine, and TMEM16A is highly expressed in ICCs (12). The results of previous studies showed that intestinal contractions in the mouse ileal segment could be inhibited by inhibiting TMEM16A in the ICCs (29). Therefore, we tested the effect of GRg2 and GRf on ileum contractions in isolated ileum tissue from guinea pigs. Acetylcholine and atropine were used to confirm the viability of the intestine before testing. The ileum exhibited a significant contraction after 50 μM GRg2 was added, and the ileum contraction was more obvious when the concentration of GRg2 was increased to 100 μM. This ileum contraction caused by GRg2 could be suppressed to baseline by 20 μM CaCC_inh-A01. GRf also produced similar effects, but the amplitude of the ileal contractions and acting concentration of GRf were different from those of GRg2 (Fig. 3 A). Statistical analysis of ileal contractile tensions and frequencies indicated that GRg2 and GRf could simultaneously increase ileal contractile tensions and frequency (Fig. 3, B and C), and this promoting action was concentration dependent.

Figure 3.

GRg2 and GRf enhance guinea pig ileum contractions. (A) acetylcholine and atropine were added and washed using Tyrode’s solution to test ileal activity (black) as control. Then, GRg2 (blue) or GRf (red) was added to the bath solution (n = 4). (B) A comparison of the contraction tension with different GRg2 and GRf concentrations (n = 4) is shown. (C) A comparison of the contraction frequencies with different GRg2 and GRf concentrations (n = 4) is shown. ^∗∗, the p-value is less than 0.01. To see this figure in color, go online.

An MTT assay was used to test the effect of GRg2 and GRf on cell viability. The results showed that GRg2 and GRf had no inhibitory effects on cell proliferation (Fig. S4). The real-time cell images in Fig. S4 B also showed that the cell morphology did not change even when the cells were incubated with 500 μM GRg2 and GRf for 24 h. Cell viability assay results showed that GRg2 and GRf were biosafety compounds, and they had no harmful effect on cell viability with concentrations 15 times or 5 times greater than the EC₅₀, respectively. Therefore, low toxicity makes it possible for GRg2 and GRf to be developed as medicine for gastrointestinal motility.

Putative binding sites between ginsenosides and TMEM16A

AutoDock 4.2 software was used to predict the putative binding sites of GRb1, GRg2, and GRf to TMEM16A. Interestingly, the results of the molecular docking indicated that all of the GRb1, GRg2, and GRf combined with the TMEM16A protein at the same amino acids locating in the transmembrane domain, including N650 and A697 of the sixth transmembrane helix, E705 of the seventh transmembrane helix, and L746 of the eighth transmembrane helix (Fig. 4 A). Fig. 4 B showed the conformation of the drug molecules and the protein at the lowest binding energy. Next, the results of the molecular docking were confirmed with site-directed mutagenesis experiments, in which N650 was mutated to alanine and E705 was mutated to glutamine and A697 was mutated to serine and L746 was mutated to threonine. Whole-cell patch-clamp experiments showed that the typical currents of both N650A/E705Q mutants and A697S/L746T mutants activated by 100 μM GRg2, 200 μM GRf, and 100 μM GRb1 were significantly reduced compared to those of the corresponding wild-type TMEM16A (Fig. 4 C). The statistical results in Fig. 4 D showed that the currents of the TMEM16A mutants activated by 100 μM GRb1, 100 μM GRg2, or 200 μM GRf were reduced more than 80% compared to those of wild-type TMEM16A. Based on these results, we confirmed that the putative binding sites of GRg2 and GRf on TMEM16A at least included N650, E705, A697, and L746.

Figure 4.

Putative binding sites between ginsenosides and TMEM16A. (A) An overall side view image of the putative binding sites showing ginsenoside molecules binding to TMEM16A. (B) A schematic diagram of the putative binding sites of ginsenosides to TMEM16A is shown. The drug molecules are indicated in black (GRb1), green (GRg2), and pink (GRf), and amino acids are indicated in yellow. (C) Typical whole-cell currents of wild-type (WT) TMEM16A, A697S/L746T mutant, and N650A/E705Q mutant activated by GRb1, GRg2, and GRf (n = 5) are shown (both the pipette solution and bath solution are free of Ca²⁺). (D) A statistical histogram of (C) is shown. (E, a–f) The typical whole-cell currents of WT TMEM16A, A697S/L746T mutant, and N650A/E705Q mutant activated by 600 nM Ca²⁺ and inhibited by 20 μM CaCC_inh-A01 are shown. (E, g and h) The typical whole-cell currents of N650A/E705Q mutant activated by 5 mM Ca²⁺ and inhibited by 20 μM CaCC_inh-A01 are shown. (E, i–l) The typical whole-cell currents of N650A/E705Q mutant activated by 600 nM Ca²⁺ or 600 nM Ca²⁺ added with 100 μΜ GRb1, 100 μΜ GRg2, and 200 μΜ GRf with membrane voltage clamped from −80 to +200 mV are shown. ^∗∗, the p-value is less than 0.01 compared with WT group. To see this figure in color, go online.

However, because N650 and E705 are also important Ca²⁺ binding sites, experiments were performed to demonstrate that the mutants still have channel function. We detected the fluorescence distribution of the cells after transfection of the double mutated TMEM16A plasmid by fluorescence experiments. The results showed that the double mutant was almost all distributed in the cell membrane, indicating that the double mutant had normal surface expression (Fig. S5 A). On the other hand, the results of Western blot experiments also showed that the protein expression of TMEM16A in cells membrane were significantly increased after transfection of the wild-type TMEM16A or the double mutant (Fig. S5 B). Therefore, we believed that the double mutant had normal surface expression. In addition, whole-cell patch clamp was performed to test the ion channel activity of the two double mutants. The results showed that the A697S/L746T mutant has had similar Ca²⁺ dependence and voltage dependence with those the wild-type TMEM16A (Fig. 4 E); however, the N650A/E705Q mutant could hardly be activated by 600 nM Ca²⁺. To demonstrate the ion channel activity of the N650A/E705Q mutant, we increased the Ca²⁺ concentration in the pipette solutions to 5 mM or set the voltage protocol of membrane voltage clamped from −80 to +200 mV in the whole-cell patch clamp (Fig. 4 E). Both the results showed that the channel currents of the N650A/E705Q mutant could be recorded. Therefore, we believe that the two double mutants still had ion channel activity. Based on the above results, we confirmed that the N650, E705, A697, and L746 are putative binding sites of GRg2 and GRf to TMEM16A, not because of losing the channel activity.

A697 and L746 bind to ginsenoside analogs through hydrophobic interactions

We extracted the conformations of different ginsenoside analogs and TMEM16A in the stable binding states. The results showed that the isobutenyl of GRb1, GRg2 and GRf were located in the middle of A697 and L746 amino acids, whereas the isobutenyl of GRb2, GRh2 and NGR1 were located in different positions (Fig. 5). On account of this particular spatial conformation, we used the Discovery Studio Visualizer software to analyze the hydrophobic interactions of these six ginsenoside analogs. The results indicated that the isobutenyl of GRb1, GRg2 and GRf bound to A697 and L746 by hydrophobic interaction, whereas the isobutenyl of GRb2, GRh2 and NGR1 were combined with amino acids other than A697 or L746, or could not form hydrophobic interactions (Fig. S6). It could be deduced that A697 and L746, which located on TM7 and TM8 of TMEM16A, could combine with the isobutenyl of ginsenoside to fix the position of ginsenoside in TMEM16A, thus affecting the opening of the TMEM16A pores. Therefore, we propose that the combination of A697 and L746 amino acids with the isobutenyl chain of ginsenosides is one of the most essential interactions for ginsenosides in activating TMEM16A.

Figure 5.

GRb1, GRg2, and GRf combine with TMEM16A by A697 and L746 through hydrophobic interaction in stable binding states. (A–F) Shown is a conformation image of GRb1, GRg2, GRf, NGR1, GRb2, GRh2 binding with A697 and L746 of TMEM16A under steady state (the isobutenyl structure of ginsenosides are pointed out in black circles). To see this figure in color, go online.

N650 and E705 bind to ginsenosides via electrostatic interaction

To explore the molecular mechanisms of the interactions between ginsenosides and TMEM16A, we calculated the surface electrostatic potential of different ginsenosides and TMEM16A. The electrostatic surface potential results showed that the putative binding sites area of the TMEM16A protein belonged to low potential area (the red region), whereas the electrostatic surface potentials of ginsenoside analogs had relatively high potentials (blue) (Fig. 6, A and C). Further results indicated that the low potential of the putative binding sites area was elevated when N650 and E705 were mutated to alanine or glutamine, respectively (Fig. 6 B). In addition, the order of the electrostatic surface potential of ginsenoside analogs from high to low were, GRg2 > GRb1 > GRf > NGR1 > GRb2 > GRh2 (Fig. 6 C), which indicated that the complementary effects of the ginsenoside analogs electrostatic potentials to TMEM16A gradually grew weaker one by one. The strength of electrostatic binding capacity of the simulation results and the activation ability of different ginsenosides in patch-clamp experiments corresponded to each other. Therefore, we concluded that the main interactions of ginsenosides with N650 and E705 were electrostatic interactions.

Figure 6.

Ginsenosides combine with TMEM16A by N650 and E705 through electrostatic interaction. (A) The electrostatic surface potential of the putative binding sites area of TMEM16A is shown. (B) N650A/E705Q mutation increases the electrostatic surface potential of the putative binding sites area in TMEM16A. The scale for electrostatic surface potential representation is red: −25 kcal/mol; white: 0 kcal/mol; and blue: +25 kcal/mol. (C) The molecular electrostatic surface potential of ginsenoside analogs is shown. The scale for electrostatic surface potential representation is red: −0.5 kcal/mol; white: 0 kcal/mol; and blue: +0.5 kcal/mol. To see this figure in color, go online.

Discussion

In this study, we explored the activation mechanism of ginsenoside analogs on TMEM16A through analysis of the interaction between six ginsenoside analogs and TMEM16A. We found that two novel ginsenoside analogs, GRg2 and GRf, can directly activate TMEM16A. Subsequently, we confirmed that GRg2 and GRf can enhance ileal contractions in guinea pigs and verified their biosafety for cell proliferation. Finally, we analyzed the molecular mechanism of the interaction between ginsenoside analogs and TMEM16A, and we proposed the gating mechanism of TMEM16A channel activated by ginsenosides.

First, we proposed the mechanism of TMEM16A channel opening process activated by ginsenosides. This TMEM16A opening mechanism activated by ginsenosides will promote the exploration of TMEM16A gating mechanism. Although the cryoEM structure of TMEM16A has been established, the research on the gating mechanism of TMEM16A is still just getting started. According to reports in the literature, the open and closed state of TMEM16A channel is mainly determined by the conformational change of transmembrane TM6-8 (8,9). In this work, our data showed that ginsenoside analogs and TMEM16A are mainly combined by electrostatic interaction (N650 on TM6 and E705 on TM7) and hydrophobic interaction (A697 on TM6 and L746 on TM8). Therefore, we proposed that the isobutenyl of ginsenoside binds to A697 on TM7 and L746 on TM8 by hydrophobic interaction, and the position of ginsenosides in TMEM16A can be fixed after binding. The ginsenoside analogs were combined with N650 on TM6 and E705 on TM7 by electrostatic interaction. Since the spatial position of ginsenoside analogs have been fixed by TM7 and TM8, and the inner half of TM6 comes into physical contact with ginsenoside analogs, the above conformational change result in the opening of the pores (Fig. 7). Both the spatial position of the isobutenyl combining with the amino acid residues (A697 and L746) and the electrostatic interaction, which leads to TM6 conformational changing, are indispensable for TMEM16A pore opening. This may be the reason why the structural similarity between GRb2 and GRb1 is more than 95%, but GRb2 still cannot activate TMEM16A. In addition, previous studies have shown that N650 and E705 were also Ca²⁺-sensitive sites (30,31). The EC₅₀ of Ca²⁺-activated wild-type TMEM16A is 0.36 μM, and the value increases to 1.8 and 231 μM when N650 and E705 are mutated to alanine and glutamine, respectively (30). Whole-cell patch-clamp results showed the EC₅₀ of Ca²⁺ was decreased when GRg2 or GRf was present in the pipette solution. Similarly, the EC₅₀ of GRg2 and GRf were also decreased when Ca²⁺ was present in the pipette solution (Fig. S7). Therefore, we concluded that the calcium ions and ginsenside analogs show a synergistic action in the activation of TMEM16A. Based on the literature and our experiment results, we confirmed that N650 and E705 are important amino acid sites in the gating mechanism of TMEM16A. However, there are more data needed to prove this result in the future.

Figure 7.

Schematic image of molecular mechanism of TMEM16A activated by ginsenosides. (left) TMEM16A in closed-state conformation. The amino acid residues N650, A697, E705, and L746 are shown with red dots. (right) TMEM16A in open-state conformation. The inner half of TM6 physically contacted with ginsenoside, which results in the opening of the pores. To see this figure in color, go online.

Second, GRg2 and GRf are effective TMEM16A activators with high specificity. The EC₅₀ of GRg2 is 32 ± 3μM, and the EC₅₀ of GRf is 101 ± 4μM (Fig. 2 E). The EC₅₀ of GRg2 is lower than that of most natural product modulators. For example, the IC₅₀ of TMEM16A natural inhibitor products eugenol, DIDS, and NPPB are 150 μM (32), 548.86, and 64.14 μM (33), respectively, and our previous study found the EC₅₀ of GRb1 is 38.4 ± 2.14 μM (11). Future work will include the modification of ginsenoside analog structures to improve efficiency. In addition, GRg2 and GRf have better specificity than other TMEM16A modulators; GRg2 and GRf can activate TMEM16A and TMEM16B but have no influence on other chloride ion channels, including bestrophin-1 and CFTR (Fig. S3). To date, almost all TMEM16A modulators have no specificity between TMEM16A and TMEM16B because of the high homology between the two proteins. However, many TMEM16A modulators have an effect on bestrophin-1, including NFA, DIDS, and CaCC_inh-A01 (33,34). The high efficiency and specificity allows GRg2 and GRf the potential to be developed as targeted drugs for TMEM16A dysfunction diseases including insufficient intestinal motility.

Third, GRg2 and GRf are relatively safe TMEM16A activators. The MTT experiments showed that even concentrations more than 15 times of the GRg2 EC₅₀ and 5 times of the GRf EC₅₀ did not significantly inhibit cell viability (Fig. S4). Furthermore, the ileal contractions after soaking in GRg2 or GRf for 20 min can be inhibited by CaCC_inh-A01. In addition, studies have shown that 80 μM GRg2 enhances cell viability and superoxide dismutase activity (14). Therefore, we determined that GRg2 and GRf are safe TMEM16A modulators, unlike other TMEM16A regulators, such as dehydroandrographolide (35), idebenone (36), and flavonoids, which have high cytotoxicity (37). On the other hand, the molecular mechanism of ginsenosides activating TMEM16A was proposed in this study. The high biosafety and clear activation mechanism make ginsenosides have great potential to be developed as medicine for diseases caused by the downregulation of TMEM16A.

In summary, GRg2 and GRf are safe and effective TMEM16A activators, and they can enhance the ileum contractions of guinea pigs by activating TMEM16A. We proposed the mechanism of TMEM16A channel opening process activated by ginsenosides. By comparing it to the process of Ca²⁺ activating TMEM16A, we conclude that N650 and E705 may be key residues for ginsenosides’ dependent activation of TMEM16A channel. The conclusion of this work will also provide important clues for the study of the TMEM16A gating mechanism.

Author Contributions

Participated in research design, S.G., Y.F.C., H.L.A., and Y.Z.; Conducted experiments, S.G. and S.S.; Performed data analysis, S.G., Y.F.C., S.S., and H.L.A.; Wrote or contributed to the writing of the manuscript, S.G., Y.F.C., S.S., C.L.P., X.Z.W., H.L.Z., H.L.A., and Y.Z.

Editor: Sudha Chakrapani.